This project will address two major bioinformatics problems: the development of new and improved software for finding genes in eukaryotic genome sequences, and the development of a sequence assembler that is capable of assembling very large genomes. The gene finding project will pursue three tracks: first, we will improve our existing eukaryotic gene finding system, GlimmerM, adding the ability to recognize new sequence patterns and enhancing the ease with which the system can be adapted to new organisms. Second, we will develop a new gene finder, based on Pair Hidden Markov Models (PHMMs), which will use the sequence similarity between two related organisms to find genes in both species simultaneously. Third, we will develop a system for integrating the output from multiple gene finders and from sequence alignment programs in order to produce gene models that incorporate all available evidence. The assembler project will include the development of several major components. The overall goal is to build a sequence assembler that will be able to assemble data from whole-genome shotgun sequencing projects for genomes ranging from a few million base pairs up to billions of base pairs. The assembler will have the ability to accept as input both raw sequencing reads and a mixture of reads and already-assembled sequences. A separate scaffold-building program will create larger scaffolds from a set of assemblies by using information from paired-end sequences. In addition, this project will develop and distribute a genome assembler benchmark set, containing sequences from shotgun sequencing projects for which the correct assembly is known. For all of the software development projects, the source code will be made freely available to investigators in the scientific research community worldwide.